Originally a dataset of 952 posts (11 variables) of the Eyewatch Ballarat facebook page were pulled on 16 May 2018.

After cleaning and tidying the data, we removed (105) posts containing ‘vicpol’ in the link. The follow graphs were created from the remaining 847 posts.

Posts by time

# A tibble: 1 x 6
  likes_mean comments_mean shares_mean mean_posts n_YearMonths n_posts
       <dbl>         <dbl>       <dbl>      <dbl>        <int>   <int>
1       26.3          4.89        21.4       24.2           34     823

There is high variability in number of posts per month. On average 24.3 posts are made per month exluding the first 4 months. Each post, on average recieves 26.3 likes, 4.9 comments and 21.4 shares.

Mean shares by time across type

Link posts tend to recieve the most shares per post, status posts use to receive more shares per post than the other types, this may be a an artifact only having a few.

Posts by hour across weekday

Posts are much more common on weekdays than weekends as expected. Posts are not lowest around early morning hours, suggesting a timezone offset. It looks like there might be a higher proportion of posts on hours “01” and “04”, suggesting a systamatic process possibly calendar event ortimed release of news.

Comments and likes by post age

Newer posts appear to receive more comments than older posts. Older posts may or may not receive for likes (hard to tell at this color and transparency)

Shares and likes by type

Extreme values tend to be photos. Extreme values tend to be liked or shared, but not both. Note that photos are the most common posts followed by links.

Like by year-month across type of post

Photos followed by links generate the most likes. Note that the y axis is likes and not likes/post.

Missingness and pair-wise scatterplot

The data is mostly present. story, message, link are the variables by descending missingness.

Metadata

page: eyewatchballarat pull time: 1.04 min raw dim: 952 obs x 11 var clean dim: 95 2obs x 26 var pull time: 5:30pm 16 May 2018